[ https://issues.apache.org/jira/browse/SPARK-13042?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15119706#comment-15119706 ]
Daniel Marcous commented on SPARK-13042: ---------------------------------------- [~srowen] Forgive me for my ignorance, read the guidelines and tried to make it right this time. Also added an update to the test file to include this metric. It seemed necessary and useful from me, I believe its an underused but important metric to take into account in some cases (including some of my own). Seemed more logical to me to include it as a metric here instead of doing it manually in user code in addition to the already existing metrics. Then again, you are the expert so I'll happily accept your decision once you get to one. > Adding Specificity metric (= true negative rate) to MulticlassMetrics.scala > --------------------------------------------------------------------------- > > Key: SPARK-13042 > URL: https://issues.apache.org/jira/browse/SPARK-13042 > Project: Spark > Issue Type: Improvement > Components: MLlib > Reporter: Daniel Marcous > Priority: Minor > Labels: metrics, mllib > Original Estimate: 12h > Remaining Estimate: 12h > > Adding Specificity metric (= true negative rate) to MulticlassMetrics.scala > 1. Add specificity calculations: > pecificity (SPC) or true negative rate > SPC = TN / N = TN / (TN+FP) > False Positive Rate (FPR) > FPR = FP / N = FP / (FP + TN) = 1-SPC > SPC = 1 - FPR > 2. Add specificity tests to : > mllib/src/test/scala/org/apache/spark/mllib/evaluation/MulticlassMetricsSuite.scala -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org